260 ROMANCE LANGUAGES 



entire knowledge of the Latin folk-speech rests on facts so meagrely 

 forthcoming or on inferences so subject to revision that it is no 

 wonder the contingent of scholars who are active in their investiga- 

 tion of it are divided into opposing camps, those who minimize 

 and those who tend to magnify the degree of difference between 

 what it is convenient to call book-Latin and the language of the 

 unlettered people; those, on the one hand, who are ready in gen- 

 eral to accept as sufficiently established well-nigh the entire series of 

 forms and phenomena deduced inferentially from the testimony 

 of the Romance languages, and those, on the other, who strenuously 

 object to this somewhat presumptuous procedure, alleging that the 

 only trustworthy data are those afforded by documentary evidence 

 of Latin origin. As to the degree of divergence between book-Latin 

 and folk-Latin, it would appear that thus far the point of view 

 has been too prevailingly that of the investigator who would fain 

 discover, boldly confronting one another, two strongly characterized 

 idioms; whereas the truer view, to be made more clear, I believe, by 

 scholars of the future, is manifestly that of a fundamental and 

 substantial unity underlying a diversity of phenomena characteristic, 

 not of two opposing modes of speech, but of a multiplicity of in- 

 fluences interacting with varying intensity among all classes of the 

 people. 



By the side of the problems of the Latin folk-speech should be 

 placed the question of the survival in the Romance languages of 

 traces of the speech of those pre-Latin races on whom the Latin was 

 imposed by conquest and colonization. Here the interesting task of 

 discovering possible indications of pre-Roman influences is rendered 

 peculiarly difficult by the tenuity of our real knowledge of the lan- 

 guages concerned; so that whether we seek for Oscan and Umbrian 

 traces in central and southern Italy, for Gallic traces in north Italy 

 and France, or for Iberian influences in the Spanish peninsula, we 

 find ourselves thrown back for the most part on scanty inferences 

 and surmises. Accordingly, notwithstanding the profound and in- 

 genious disquisitions on this subject, conducted chiefly by Ascoli 

 and by Meyer-Liibke, it still remains delightfully problematical 

 whether even so generally supposed a Celtic trait as the li-sound 

 for Latin long u prevailing throughout north and south France, 

 Piedmont, Genoa, and Lombardy, and in the direction of the Grisons 

 and the Tyrol, is certainly to be attributed to such a source. For 

 the nasal vowels, which appear over approximately the same territory 

 with the addition of Portugal, the probability of primitive Celtic 

 influence is perhaps somewhat more assured. As to the possible 

 traces of Oscan and Umbrian in Italian, two or three consonantal 

 developments are all that in the present state of knowledge can 

 be referred to those dialects; while for the Iberian influence on the 



